How Canberra's Public Records Ended Up Full of Ghost Images: The Long Road to Duplicate Image Replacement

ACT government databases and federal agency document repositories hold tens of thousands of duplicate image files — scanned records, identification photos, and infrastructure diagrams stored multiple times across overlapping systems — and the effort to clean them up has become one of the quietest, most persistent headaches in Canberra's sprawling public service machinery.

The issue matters now because several large-scale digitisation programs are converging at once. The National Archives of Australia, based in Parkes, is midway through an accelerated digitisation push. The ACT government's own Shared Services ICT division has been consolidating agency databases following the machinery-of-government changes that accompanied the post-2022 ACT election restructure. Both processes have surfaced the same problem: records migrated from legacy systems brought their duplicates with them, compounding over years of piecemeal scanning projects that lacked a common deduplication standard.

How the Backlog Built Up

The roots go back to the early 2000s, when individual ACT directorates began scanning paper files independently, with no central authority setting file-naming conventions or mandating hash-based deduplication checks. The Australian Public Service Commission's 2019 capability review of digital record management flagged fragmentation across agencies as a systemic risk, but resourcing to address it remained inconsistent through successive budget cycles.

When the ACT government rolled out its Digital Services Renewal program from 2021 onward, teams at Civic's Canberra Nara Centre and within the Transport Canberra and City Services directorate on Callam Street in Phillip discovered that migrating records to new content management platforms was surfacing duplicate image counts far higher than anticipated. Infrastructure project files — roads work around Flemington Road in Gungahlin, light rail corridor surveys along Northbourne Avenue — had been scanned by multiple teams at different stages of approval, then uploaded without cross-referencing existing holdings.

University of Canberra researchers working with the ACT government under a 2023 data governance partnership estimated that duplicate image files in mid-sized government content repositories typically account for between 12 and 18 percent of total stored volume, though the ACT government has not published its own agency-specific figures. At current Amazon Web Services pricing benchmarks for government cloud storage tiers — roughly $0.025 per gigabyte per month at standard rates — that redundancy translates to a measurable ongoing cost at scale, though precise ACT expenditure figures are not publicly available.

The Push to Fix It

The practical response has taken two broad forms. First, automated deduplication tools — software that identifies files sharing identical pixel-level content regardless of filename — have been introduced into the migration pipeline at several ACT directorates since mid-2024. Second, the ACT's digital records policy, last comprehensively updated in 2018, is under review by the Office of the Chief Digital Officer, with a revised framework expected to be tabled before the ACT Legislative Assembly's administration committee later this year.

For frontline public servants in Belconnen offices and along Constitution Avenue, the practical effect has been disruption during database migrations, with staff occasionally unable to locate the canonical version of a document when multiple scans appear in search results. Training programs run through the Australian Public Service Commission's LearnHub platform have been updated to include guidance on image file provenance, though uptake tracking has not been made public.

What happens next depends substantially on whether the revised digital records policy mandates deduplication as a condition of any new scanning contract, rather than leaving it to agency discretion. Procurement officers at the Department of Finance's Deloitte Tower offices in the city have been working through updated ICT procurement guidelines that would embed that requirement from the tender stage. If the policy lands before the end of 2026, the next round of archiving contracts — expected to be let in the first quarter of 2027 — would be the first to carry the obligation as a hard contractual term, not a recommendation. For anyone managing records in the capital, that distinction is not small.